Feedback on Átila's Vision for D
welkam
wwwelkam at gmail.com
Thu Oct 17 15:26:15 UTC 2019
On Tuesday, 15 October 2019 at 13:09:15 UTC, Mike Parker wrote:
> This thread is for general feedback and discussion regarding
> Átila's blog post on his vision for D's future.
>
> https://dlang.org/blog/2019/10/15/my-vision-of-ds-future/
This whole thing reads more like what Átila wants and reasonably
could do in a year than a vision for D. Its more like I instead
of we. There is effort to port D to android and I want to see
that discoursed in a vision document. What is currently is being
done. When we expect what to be done. What needs to be done but
needs more funding. Those kind of things. I want to hear more
about the whole effort not only yours. I want to see a plan and a
vision.
In that document you talked about making D also an interpreted
language. I dont think its that good of an idea. While I would
love to have faster compile times I think its better to spend
time optimizing DMD. While it is fast game developers would not
call it well optimized. If we applied data oriented design ideas
to both front end and back end that would benefit everyone.
Second crazy idea I have is to merge lld and DMD together. Now
compiler translates internal data structures to object file and
then linker read the file and tranlate it to internal structures.
With modifications to both DMD and lld its possible to go make
lld to just read DMD internal data. That would speed up build
times.
The other thing that is completely missing from any discussion is
how to improve onboarding process for new people that want to
work on compiler, runtime or phobos. There is little
documentation on them. Its not clear where to even ask questions
about them. There is no help in "breaking the ice" so that new
people become contributors. Personally I wanted to dump the whole
data structure that is build here
https://github.com/dlang/dmd/blob/master/src/dmd/attrib.d#L133
After some time I failed to find a place where its completely
built up and where could I inspect it. I failed because I jumped
in the middle of compiler without knowing the big picture view on
how compiler works. Then I decided that I would read from main to
understand how things work. I reached tokenizer and abandoned the
whole idea. If a person have such a problem should they ask
Walter or maybe Sebastian Wilzbach? And if so should they write
private emails? At the time I was looking at compiler code I
didnt knew that there were such a thing as slack. But it requires
an invite and I dont have it.
There are many problems preventing people from contributing and
there is not even a discussion about improvements to the process.
Also I would like to mention that spending time attracting people
might be more useful than spending that time coding.
More information about the Digitalmars-d
mailing list